GPT 5.4 Nano vs Inkling: Benchmark Comparison
GPT 5.4 Nano has the higher score on 4 of 18 shared benchmarks; Inkling leads on 14.
The largest observed score gap is 22.11 pts on Legal Research Bench , where Inkling leads.
Reported ±1 standard-error ranges overlap on 6 of 18 shared benchmarks with comparable uncertainty data. This is not a pairwise statistical significance test.
Shared benchmark results
Scores and reported standard errors come from the latest visible Vals benchmark version available for both models.
| Benchmark | GPT 5.4 Nano | Inkling | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 0.00% ±0.00 | 2.08% ±0.83 | 2.08 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 6.25% ±1.68 | 28.36% ±3.13 | 22.11 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 77.92% ±0.43 | 82.95% ±0.50 | 5.03 pts | Reported ±1 SE ranges do not overlap |
| EMB | 44.75% ±3.12 | 38.36% ±3.27 | 6.39 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 38.22% ±1.19 | 46.60% ±0.93 | 8.38 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 59.10% ±0.96 | 63.99% ±0.94 | 4.89 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 26.58% ±2.52 | 43.52% ±2.95 | 16.94 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 67.42% ±0.93 | 75.31% ±0.84 | 7.89 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 41.03% ±2.26 | 41.19% ±2.23 | 0.16 pts | Reported ±1 SE ranges overlap |
| MedScribe | 77.09% ±1.89 | 85.41% ±1.84 | 8.31 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 77.53% ±2.42 | 87.12% ±1.78 | 9.60 pts | Reported ±1 SE ranges do not overlap |
| MMLU Pro | 77.17% ±0.43 | 86.30% ±0.34 | 9.12 pts | Reported ±1 SE ranges do not overlap |
| MMMU Pro | 73.58% ±1.06 | 78.50% ±0.99 | 4.91 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 38.08% ±3.09 | 36.55% ±3.22 | 1.53 pts | Reported ±1 SE ranges overlap |
| Code Migration | 14.47% ±4.03 | 11.79% ±3.40 | 2.68 pts | Reported ±1 SE ranges overlap |
| LiveCodeBench | 84.01% ±1.04 | 85.52% ±1.00 | 1.51 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 69.80% ±2.06 | 77.60% ±1.87 | 7.80 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 26.10% ±5.08 | 19.21% ±3.66 | 6.88 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | GPT 5.4 Nano average | Inkling average |
|---|---|---|
| Legal | 28.06% | 37.80% |
| Finance | 47.21% | 53.55% |
| Healthcare | 59.06% | 63.30% |
| Academic | 76.09% | 83.97% |
| Education | 38.08% | 36.55% |
| Coding | 48.59% | 48.53% |
Cost and latency
Cost per test appears only where the benchmark reports it for both models. Latency is the measured completion time for that benchmark.
| Benchmark | GPT 5.4 Nano cost | Inkling cost | GPT 5.4 Nano latency | Inkling latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $0.18 | $1.25 | 6m03s | 18m24s |
| Legal Research Bench | $0.14 | $0.49 | 8m47s | 14m29s |
| LegalBench | N/A | N/A | 2.70s | 23.14s |
| EMB | $0.57 | $1.67 | 29m27s | 26m22s |
| Finance Agent (v2) | $0.16 | $1.00 | 5m35s | 18m02s |
| MortgageTax | N/A | N/A | 14.53s | 69.11s |
| Tax Agent Bench | $0.07 | $0.35 | 3m40s | 13m31s |
| TaxEval v2 | N/A | N/A | 15.90s | 3m18s |
| MedCode | N/A | N/A | 8.04s | 2m47s |
| MedScribe | N/A | N/A | 20.53s | 4m05s |
| GPQA Diamond | N/A | N/A | 29.50s | 5m09s |
| MMLU Pro | N/A | N/A | 8.21s | 73.26s |
| MMMU Pro | N/A | N/A | 21.91s | 2m04s |
| SAGE | N/A | N/A | 34.33s | 3m18s |
| Code Migration | $0.42 | $3.01 | 26m34s | 29m34s |
| LiveCodeBench | N/A | N/A | 60.09s | 4m18s |
| SWE-bench | $0.10 | $0.59 | 4m23s | 6m34s |
| Vibe Code Bench v1.1 | $1.28 | $1.61 | 54m36s | 18m25s |
Results available only for GPT 5.4 Nano
None.
Results available only for Inkling
- Vals Index
- ProofBench v1.1
- IOI
- ProgramBench
- SkillsBench
- Terminal-Bench 4.0
- Vibe Code Bench 1-100
- CyberBench v1.1
- Public Benefits Bench v1.1